Arch Linux and Sage
Josh
sage: An error occurred while installing sqlite-3.6.17
Please email sage-devel http://groups.google.com/group/sage-devel
explaining the problem and send the relevant part of
of /home/josh/Downloads/sage-4.2/install.log. Describe your computer,
operating system, etc.
If you want to try to fix the problem yourself, *don't* just cd to
/home/josh/Downloads/sage-4.2/spkg/build/sqlite-3.6.17 and type
'make'.
Instead type "/home/josh/Downloads/sage-4.2/sage -sh"
in order to set all environment variables correctly, then cd to
/home/josh/Downloads/sage-4.2/spkg/build/sqlite-3.6.17
(When you are done debugging, you can type "exit" to leave the
subshell.)
make[1]: *** [installed/sqlite-3.6.17] Error 1
make[1]: Leaving directory `/home/josh/Downloads/sage-4.2/spkg'
real 0m0.774s
user 0m0.723s
sys 0m0.043s
Error building Sage.
I installed sqlite, is there still stuff that is wrong?
Tim Joseph Dumol
Currently, Sage does not build properly in Arch Linux. You may want to install the binary version at http://sagemath.org or at the Arch Linux AUR ( http://trac.sagemath.org/sage_trac/ticket/7408 ).
- Tim Joseph Dumol <tim (at) timdumol (dot) com>
http://timdumol.com
Josh
On Nov 7, 8:54 pm, Tim Joseph Dumol <t...@timdumol.com> wrote:
> Hey Josh,
>
> Currently, Sage does not build properly in Arch Linux. You may want to
>
> > So when attempting to run make I got this:
>
> > sage: An error occurred while installing sqlite-3.6.17
Tim Joseph Dumol
* Incompatible libreadline.so.6 -- the libreadline in Sage is dynamically linked to libtermcap. Arch Linux disables libtermcap, so one needs to copy the system's libreadline to $SAGE_ROOT/local/lib. This is also the case in binary distributions.
* Issue building R -- this is a problem in many platforms with up-to-date GCC packages, and is worked around by either copying the system's libgcc, or by building Sage with the environment vars SAGE_FORTRAN and SAGE_FORTRAN_LIB set to the system's gfortran and libgfortran.so.
* Build hangs while building libfplll, which is worked around by copying the system's libfplll, and editing the $SAGE_ROOT/local/installed/ to make the build system think that libfplll is installed.
* Finally, after the build is done, starting Sage or running the doctests fails with a message about a function in the Singular interface not being implemented. I do not know any workarounds to this.
Thus, it would probably be best to use the binary distribution for now.
William Stein
> The main issues are:
>
> * Incompatible libreadline.so.6 -- the libreadline in Sage is dynamically
> linked to libtermcap. Arch Linux disables libtermcap, so one needs to copy
> the system's libreadline to $SAGE_ROOT/local/lib. This is also the case in
> binary distributions.
Maybe this could be fixed by not linking to termcap on archlinux.
This would just require changing the spkg-install.
> * Issue building R -- this is a problem in many platforms with up-to-date
> GCC packages, and is worked around by either copying the system's libgcc, or
> by building Sage with the environment vars SAGE_FORTRAN and SAGE_FORTRAN_LIB
> set to the system's gfortran and libgfortran.so.
We will switch to requiring gfortran by default soon. For now do
export SAGE_FORTRAN=/usr/bin/gfortran
export SAGE_FORTRAN_LIB="/usr/lib/libgfortran.so"
before building sage.
> * Build hangs while building libfplll, which is worked around by copying the
> system's libfplll, and editing the $SAGE_ROOT/local/installed/ to make the
> build system think that libfplll is installed.
I didn't see this at all.
> * Finally, after the build is done, starting Sage or running the doctests
> fails with a message about a function in the Singular interface not being
> implemented. I do not know any workarounds to this.
I didn't see this at all either.
>
> Thus, it would probably be best to use the binary distribution for now.
>
I've only built a 64-bit binary distribution, since I haven't setup a
32-bit archlinux box yet. I will, but not now, since it takes a
while (even given the 32-bit install Dan Drake gave me -- there's
still a lot of config work I have to do).
SUMMARY: I built the 64-bit binary. For me Sage starts fine and
*all* doctests pass (both normal and long), so I don't at all see the
Singular issues you mention. I also definitely didn't have any
trouble with libfplll. Is there really a system libfplll on your
system? That seems unlikely, given that it is such an incredibly
specialized library for a specific part of math.
Anyway, I want to emphasize that the officially supported Linux
distros for building Sage are:
Debian, Ubuntu, CentOS (=Redhat), Fedora Core, OpenSuse, Mandriva
It would be nice to add arch linux to the list, but we're not quite
there yet. I'm confident we will be soon.
William
-- William
Alex Ghitza
I've been building Sage on 32-bit Archlinux for at least one year now.
With 4.2, I only have two problems: the sqlite issue and the R issue.
Thanks to previous threads on this sort of thing, I work around them
by copying the system libreadline.so and libgcc.so to the appropriate
places in the sage build directory.
It's a bit annoying because as far as I can tell I have to wait for
sqlite to fail before copying libreadline, and then for R to fail
before copying libgcc. So the whole thing needs a bit too much of my
attention.
On the other hand, after that I have sage-4.2-goodness with longtests
passing. There are no problems with Singular on my system.
Best,
Alex
On Sat, Nov 07, 2009 at 09:06:36PM -0800, William Stein wrote:
>
> > * Build hangs while building libfplll, which is worked around by copying the
> > system's libfplll, and editing the $SAGE_ROOT/local/installed/ to make the
> > build system think that libfplll is installed.
>
> I didn't see this at all.
>
> > * Finally, after the build is done, starting Sage or running the doctests
> > fails with a message about a function in the Singular interface not being
> > implemented. I do not know any workarounds to this.
>
> I didn't see this at all either.
>
[...]
--
Alex Ghitza -- Lecturer in Mathematics -- The University of Melbourne
-- Australia -- http://www.ms.unimelb.edu.au/~aghitza/
William Stein
>
>
> I've been building Sage on 32-bit Archlinux for at least one year now.
>
> With 4.2, I only have two problems: the sqlite issue and the R issue.
> Thanks to previous threads on this sort of thing, I work around them
> by copying the system libreadline.so and libgcc.so to the appropriate
> places in the sage build directory.
>
> It's a bit annoying because as far as I can tell I have to wait for
> sqlite to fail before copying libreadline, and then for R to fail
> before copying libgcc. So the whole thing needs a bit too much of my
> attention.
You should set SAGE_FORTRAN to completely avoid the R issue.
> On the other hand, after that I have sage-4.2-goodness with longtests
> passing. There are no problems with Singular on my system.
Cool -- I'm glad the problem isn't 32-bit only (say).
-- William
Tim Joseph Dumol
Perhaps I'll try reformatting my system sometime to see if it's a problem with the libraries on my system.
Josh
William Stein
>
> What is your work around for the sqlite issue?
system-wide readline library into SAGE_ROOT/local/lib/, as you
suggested.
William
William Stein
Associate Professor of Mathematics
University of Washington
http://wstein.org
Josh
compile.
On Nov 8, 6:39 pm, William Stein <wst...@gmail.com> wrote:
>
> > What is your work around for thesqliteissue?
>
> I don't know who you're asking, but for me it's to copy the
> system-wide readline library into SAGE_ROOT/local/lib/, as you
> suggested.
>
> William
>
>
>
> > On Nov 8, 12:13 am, Alex Ghitza <aghi...@gmail.com> wrote:
> >> I've been building Sage on 32-bit Archlinux for at least one year now.
>
> >> by copying the system libreadline.so and libgcc.so to the appropriate
> >> places in the sage build directory.
>
> >> It's a bit annoying because as far as I can tell I have to wait for
William Stein
>
> I was wondering what I am suppose to do with libreadline to get it to
> compile.
Quoting from Tim's email: "* Incompatible libreadline.so.6 -- the
Alex Ghitza
Yep. On my system, when I get the sqlite error, I type
cp /usr/lib/libreadline.so /opt/sage-4.2/local/lib/
and then make again.
Best,
Alex
Josh
I just haven't found any problems, so I'm assuming its great.
On Nov 8, 7:00 pm, Alex Ghitza <aghi...@gmail.com> wrote:
> On Sun, Nov 08, 2009 at 03:57:12PM -0800, William Stein wrote:
>
Alex Ghitza
>
> Thank you! This appears to have worked great!
> I just haven't found any problems, so I'm assuming its great.
>
There's a way to check this, if you want to be sure.
Go into the directory of your freshly-installed Sage, and type
make test
This runs most of the Sage test suite (it could take a while,
depending on your machine) and reports on failures. There is also a
longer version, unsurprisingly called
make testlong
which runs the whole test suite.
Dr. David Kirkby
> We will switch to requiring gfortran by default soon. For now do
>
> export SAGE_FORTRAN=/usr/bin/gfortran
> export SAGE_FORTRAN_LIB="/usr/lib/libgfortran.so"
What happens if someone wants to use another compiler, other than GCC? I'm
thinking of Sun Studio of course.
dave
Josh
process. Is this why I would have had to copy libgcc.so to the
appropriate directory and if so, what directory?
Alex Ghitza
>
> I got the error when it attempted to compile R, 113 minutes into the
> process. Is this why I would have had to copy libgcc.so to the
> appropriate directory and if so, what directory?
>
Here is where you have two options:
1. As William wrote in a previous message, do
export SAGE_FORTRAN=/usr/bin/gfortran
export SAGE_FORTRAN_LIB="/usr/lib/libgfortran.so"
This assumes you have gfortran installed on your system. If not, do
pacman -S gcc-fortran
Now if you run "make" there should be no problems with R any more.
(Note that I haven't tried this option yet, but it should be the right
way to go. It also has the advantage that you can do it at the very
beginning, instead of having to wait until R fails.)
2. The other option (more of a hackish workaround) is to do
cp /usr/lib/libgcc_s.so /opt/sage-4.2/local/lib/gcc-lib/i686-pc-linux-gnu/4.0.3/
and try "make" again.
William Stein
Give the Fortran that works with that compiler instead of gfortran.
There's nothing GCC specific about the environment variables
SAGE_FORTRAN and SAGE_FORTRAN_LIB.
William
Dr. David Kirkby
OK,
I mis-understood your use of "requiring gfortran".
Dave